Developing professionally oriented communicative competence: Evaluating the factors of integrating large language models into foreign language instruction of non-linguistic university students

Introduction. The main research problem of the article lies in the key contradiction between the high technological potential of ChatGPT for foreign language teaching and learning and the significant risks associated with its integration, especially in the context of non-linguistic universities. Resolving this contradiction requires establishing a pedagogical balance through controlled interaction with artificial intelligence. This would allow to minimize associated risks and direct the potential of large language models like ChatGPT towards the development of professionally oriented communicative competence rather than abstract language skills. The aim of this study is to identify, summarize, and comprehensively assess key factors affecting the integration of ChatGPT into foreign language instruction of non-linguistic university students. In order to achieve this aim, the study also included an analysis of both the restraining barriers related to the technology’s use and the conditions necessary for its pedagogically-based and effective integration into the educational process. Materials and Methods. The research methodology was based on the synthesis of three complementary components: 1) conceptual (assessment of students’ perceptions of ChatGPT within the Technology Acceptance Model, TAM); 2) analytical (structuring the potential and risks of integration using SWOT analysis); 3) design-oriented (modeling an integration scenario based on the pedagogical strategy “Think–Talk–Write” (TTW). Among the major empirical methods employed were mathematical methods of statistics, which included the collection, analysis (both qualitative and quantitative), and interpretation of the data obtained from an anonymous survey of 113 students with varying language proficiency levels. During the study, the authors developed a specialized questionnaire, consisting of two blocks of closed-ended, semi closed-ended, and open-ended questions aimed at identifying behavioral patterns of ChatGPT usage and students’ subjective attitudes towards its integration into foreign language. The design of a potential integration scenario was carried out through conceptual modeling built on the theoretical analysis (TAM model, SWOT analysis) and the empirical survey (questionnaire). Results. Based on empirical data, the following results were obtained: 1. Taking into account characteristic features and subjective assessment of interaction with ChatGPT for educational purposes, the study identified behavioral patterns of the technology’s usage and its perceptions among non-linguistic university students. 2. The TAM model revealed high perceived usefulness and ease of use of ChatGPT, as well as a number of ethical and psychological barriers hindering its full integration. 3. The SWOT analysis empirically confirmed a complex of factors promoting or restraining the integration of ChatGPT into foreign language instruction of non-linguistic university students. 4. In response to the key contradiction between the enormous practical potential of the technology for solving specific learning tasks and its uncritical use, a pedagogical model of integration based on the TTW strategy was developed. 5. Conditions for implementing this model were determined, which are driven by the need to address primary language learning problems in non-linguistic universities; a specific mechanism for overcoming these problems through controlled integration of artificial intelligence into blended learning was proposed. Conclusions. The study concludes that the successful integration of ChatGPT requires the development of conceptually new models that would balance the technology’s potential with the necessity of strict pedagogical control. The theoretically and empirically justified model proposed in this study not only acknowledges the need for such control but also provides an embedded mechanism for its implementation within the educational process. The model demonstrates that overcoming risks of passivity, superficial learning, and academic dishonesty of non-linguistic university students is possible not through abandoning the technology but through its controlled integration into blended learning structure. According to the authors, such balanced yet controllable interaction with ChatGPT opens the way to risk reduction and the unlocking of the technology’s potential, thereby ensuring a high-quality digital transformation of education.